File size: 3,810 Bytes
5197190 dd80869 0382071 32bdce1 ff5de19 5197190 638960a d8c82ac 638960a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 |
---
language: zh
widget:
- text: "[CLS] 万 叠 春 山 积 雨 晴 ，"
- text: "[CLS] 青 山 削 芙 蓉 ，"
---
# Chinese GPT2 Language Models
## Model description
This is the set of two Chinese GPT2 language models pre-trained by [UER-py](https://www.aclweb.org/anthology/D19-3041.pdf).
You can download the two Chinese GPT2 language models via HuggingFace from the links below:
| Model | [gpt2-chinese-poem][poem] | [gpt2-chinese-couplet][couplet] |
| :-----------: | :------------------------------------------: | :-------------------------------------: |
| Training data | Contains about 800,000 chinese ancient poems | Contains about 700,000 chinese couplets |
## How to use
Because the parameter ***skip_special_tokens*** is used in the ***pipelines.py*** , special tokens such as [SEP], [UNK] will be deleted, and the output results may not be neat.
You can use this model directly with a pipeline for text generation:
When the parameter ***skip_special_tokens*** is True:
```python
>>> from transformers import BertTokenizer, GPT2LMHeadModel, TextGenerationPipeline
>>> from transformers import TextGenerationPipeline,
>>> tokenizer = BertTokenizer.from_pretrained("uer/gpt2-chinese-poem")
>>> model = GPT2LMHeadModel.from_pretrained("uer/gpt2-chinese-poem")
>>> text_generator = TextGenerationPipeline(model, tokenizer)
>>> text_generator("[CLS]梅 山 如 积 翠 ，", max_length=50, do_sample=True)
[{'generated_text': '[CLS]梅 山 如 积 翠 ， 的 手 堪 捧 。 遥 遥 仙 人 尉 ， 盘 盘 故 时 陇 。 丹 泉 清 可 鉴 ， 石 乳 甘 于 。 行 将 解 尘 缨 ， 于 焉 蹈 高 踵 。 我'}]
```
When the parameter ***skip_special_tokens*** is Flase:
```python
>>> from transformers import BertTokenizer, GPT2LMHeadModel, TextGenerationPipeline
>>> from transformers import TextGenerationPipeline,
>>> tokenizer = BertTokenizer.from_pretrained("uer/gpt2-chinese-poem")
>>> model = GPT2LMHeadModel.from_pretrained("uer/gpt2-chinese-poem")
>>> text_generator = TextGenerationPipeline(model, tokenizer)
>>> text_generator("[CLS]梅 山 如 积 翠 ，", max_length=50, do_sample=True)
[{'generated_text': '[CLS]梅 山 如 积 翠 ， 的 [UNK] 手 堪 捧 。 遥 遥 仙 人 尉 ， 盘 盘 故 时 陇 。 丹 泉 清 可 鉴 ， 石 乳 甘 可 捧 。 银 汉 迟 不 来 ， 槎 头 欲 谁 揽 。 何'}]
```
## Training data
Contains about 800,000 chinese ancient poems.
## Training procedure
Models are pre-trained by [UER-py](https://github.com/dbiir/UER-py/) on [Tencent Cloud TI-ONE](https://cloud.tencent.com/product/tione/). We pre-train 200,000 steps with a sequence length of 128.
```
python3 preprocess.py --corpus_path corpora/poem.txt \
--vocab_path models/google_zh_vocab.txt \
--dataset_path poem.pt --processes_num 16 \
--seq_length 128 --target lm
```
```
python3 pretrain.py --dataset_path poem.pt \
--vocab_path models/google_zh_vocab.txt \
--output_model_path models/poem_gpt_base_model.bin \
--config_path models/bert_base_config.json --learning_rate 5e-4 \
--tie_weight --world_size 8 --gpu_ranks 0 1 2 3 4 5 6 7 \
--batch_size 64 --report_steps 1000 \
--save_checkpoint_steps 50000 --total_steps 200000 \
--embedding gpt --encoder gpt2 --target lm
```
### BibTeX entry and citation info
```
@article{zhao2019uer,
title={UER: An Open-Source Toolkit for Pre-training Models},
author={Zhao, Zhe and Chen, Hui and Zhang, Jinbin and Zhao, Xin and Liu, Tao and Lu, Wei and Chen, Xi and Deng, Haotang and Ju, Qi and Du, Xiaoyong},
journal={EMNLP-IJCNLP 2019},
pages={241},
year={2019}
}
```
[poem]: https://huggingface.co/uer/gpt2-chinese-poem
[couplet]: https://huggingface.co/uer/gpt2-chinese-couplet |